# Copyright 2022-2023 OmniSafe Team. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
"""Implementation of the P3O algorithm."""
import torch
import torch.nn.functional as F
from torch.nn.utils.clip_grad import clip_grad_norm_
from omnisafe.algorithms import registry
from omnisafe.algorithms.on_policy.base.ppo import PPO
from omnisafe.utils import distributed
[docs]@registry.register
class P3O(PPO):
"""The Implementation of the P3O algorithm.
References:
- Title: Penalized Proximal Policy Optimization for Safe Reinforcement Learning
- Authors: Linrui Zhang, Li Shen, Long Yang, Shixiang Chen, Bo Yuan, Xueqian Wang, Dacheng Tao.
- URL: `P3O <https://arxiv.org/pdf/2205.11814.pdf>`_
"""
[docs] def _init_log(self) -> None:
"""Log the P3O specific information.
+-------------------+-----------------------------------+
| Things to log | Description |
+===================+===================================+
| Loss/Loss_pi_cost | The loss of the cost performance. |
+-------------------+-----------------------------------+
"""
super()._init_log()
self._logger.register_key('Loss/Loss_pi_cost', delta=True)
[docs] def _loss_pi_cost(
self,
obs: torch.Tensor,
act: torch.Tensor,
logp: torch.Tensor,
adv_c: torch.Tensor,
) -> torch.Tensor:
r"""Compute the performance of cost on this moment.
We compute the loss of cost of policy cost from real cost.
.. math::
L = \mathbb{E}_{\pi} \left[
\frac{\pi^{'} (a|s)}{\pi (a|s)} A^{C}_{\pi_{\theta}} (s, a)
\right]
where :math:`A^{C}_{\pi_{\theta}} (s, a)` is the cost advantage, :math:`\pi (a|s)` is the
old policy, and :math:`\pi^{'} (a|s)` is the current policy.
Args:
obs (torch.Tensor): The ``observation`` sampled from buffer.
act (torch.Tensor): The ``action`` sampled from buffer.
logp (torch.Tensor): The ``log probability`` of action sampled from buffer.
adv_c (torch.Tensor): The ``cost_advantage`` sampled from buffer.
Returns:
The loss of the cost performance.
"""
self._actor_critic.actor(obs)
logp_ = self._actor_critic.actor.log_prob(act)
ratio = torch.exp(logp_ - logp)
surr_cadv = (ratio * adv_c).mean()
Jc = self._logger.get_stats('Metrics/EpCost')[0] - self._cfgs.algo_cfgs.cost_limit
loss_cost = self._cfgs.algo_cfgs.kappa * F.relu(surr_cadv + Jc)
self._logger.store({'Loss/Loss_pi_cost': loss_cost.mean().item()})
return loss_cost.mean()
[docs] def _update_actor(
self,
obs: torch.Tensor,
act: torch.Tensor,
logp: torch.Tensor,
adv_r: torch.Tensor,
adv_c: torch.Tensor,
) -> None:
"""Update policy network under a double for loop.
The pseudo code is shown below:
.. code-block:: python
for _ in range(self.cfgs.actor_iters):
for _ in range(self.cfgs.num_mini_batches):
# Get mini-batch data
# Compute loss
# Update network
.. warning::
For some ``KL divergence`` based algorithms (e.g. TRPO, CPO, etc.), the ``KL divergence``
between the old policy and the new policy is calculated. And the ``KL divergence`` is
used to determine whether the update is successful. If the ``KL divergence`` is too
large, the update will be terminated.
Args:
obs (torch.Tensor): ``observation`` stored in buffer.
act (torch.Tensor): ``action`` stored in buffer.
logp (torch.Tensor): ``log_p`` stored in buffer.
adv_r (torch.Tensor): ``reward_advantage`` stored in buffer.
adv_c (torch.Tensor): ``cost_advantage`` stored in buffer.
"""
loss_reward = self._loss_pi(obs, act, logp, adv_r)
loss_cost = self._loss_pi_cost(obs, act, logp, adv_c)
loss = loss_reward + loss_cost
self._actor_critic.actor_optimizer.zero_grad()
loss.backward()
if self._cfgs.algo_cfgs.use_max_grad_norm:
clip_grad_norm_(
self._actor_critic.actor.parameters(),
self._cfgs.algo_cfgs.max_grad_norm,
)
distributed.avg_grads(self._actor_critic.actor)
self._actor_critic.actor_optimizer.step()